Definites, Domain Restriction, and Discourse Structure in Online Processing

Abstract

Definite descriptions are commonly assumed to involve a uniqueness requirement, which is crucially constrained by contextual domain restriction. Theoretical proposals differ with regards to whether a variable for domain restriction should be represented in the linguistic representation or not, and if so, whether it should be seen as contributing a property or a situation. From the perspective of actual language use and comprehension, a key question is just how contextual information is integrated for purposes of domain restriction. Two visual world eye tracking studies addressing these issues are presented. They look at participants’ eye movements as they visually inspect an array of colored shapes and listen to descriptions thereof. For example, ‘The circle is black’ is evaluated relative to a display that contains two circles in different colors and positions. This is preceded by a context sentence that helps to set up a domain that narrows the referential choice to varying degrees, e.g. by containing ‘on the top.’ Various measures are used to assess to what extent the circle that happens to be at the top is taken to be the referent of the definite description, both in real time online while the sentence unfolds and in terms of ultimate response behavior. The results suggest that people are very much sensitive to the subtle contextual clues, and in particular that the discourse status of the key prepositional phrase in the discourse context is crucial. This has implications for theoretical perspectives on domain restriction, based on their capability to incorporate the role of discourse structure.